I performed the following procedure because it appeared the shortest and least error-prone way to accomplish the replacement of libpcre.a



I got the same results. Is there anything wrong with my procedure?

I have a solution.

The problem is the the link order is fetching regcomp and regexec from libc and not from pcre. I did this:

Then I hacked the output like so:

Notice that I also edited various paths, but the big change was to include the pcre stuff on the command line so it would be linked first.

Is this helpful?

seems I have to update configure to get local pcre not involved if POSIX-compatible regXXX functions are available from libc.

My experience and my reading of other posts leads me to conclude that the problem is not dbg-cli supplied pcre v. local pcre, but mixing them. I used the bdg-cli supplied code exclusively and I saw a post of somebody that uesd the locally supplied. Both of us succeeded.

Apparently EITHER will work properly, but mixing the dbg-cli supplied header with the locally supplied (libc) code seems to be catrostrophic. I support this by observing that the segfault happend while unwinding the stack, so something stored on the stack in the auto variables was mishandled.

Now that I have mine working I just need to figure out what to do with it ...

you can step code, set breakpoints, inspect variables...
in general, that's all...